CN1342354A - Combined channel coding and space-block coding in multi-antenna arrangement - Google Patents

Abstract

Enhanced performance is achieved by combining channel coding with the space-time coding principles. With K synchronized terminal units transmitting on N antennas to a base station having M DOLLAR m(G K) receive antennas, increased system capacity and improved performance are attained by using a concatenated coding scheme where the inner code is a space-time block code and the outer code is a conventional channel error correcting code. Information symbols are first encoded using a conventional channel code, and the resulting signals are encoded using a space-time block code. At the receiver, the inner space-time block code is used to suppress interference from the other co-channel terminals and soft decisions are made about the transmitted symbols. The channel decoding that follows makes the hard decisions about the transmitted symbols. Increased data rate is achieved by, effectively, splitting the incoming data rate into multiple channels, and each channel is transmitted over its own terminal.

Description

Combination of channel coding and space-block coding in a multi-antenna arrangement

The present invention relates to wireless communications, and more particularly to techniques for efficient wireless communications in the presence of fading, co-channel interference, and other degrading effects.

The physical limitations of wireless channels pose important technical challenges for reliable communication. Bandwidth limitations, propagation loss, timing skew, noise, interference, and multipath fading make wireless channels a narrow "pipe" through which data streams cannot easily pass. In addition, other challenges arise from power constraints, size, and speed of devices used within portable wireless devices.

Using multiple transmit antennas at both the base station and the remote station can increase the wireless channel capacity, and information theory can measure this increased capacity. A standard solution to exploit this capacity is the linear processing of receivers, which is described, for example, in J. Winters, J. Salz and R.D. Gitlin, in "The impact of antenna diversity an the capacity of wireless communication systems" IEEE Trans. Communications, Vol. 42. No. 2/3/4, pp. 1740-1751, described in Feb/March/April 1994. Wittneben in "Base station modulation diversity for digital SIMULCAST", proc.IEEE'VTC, pp.505-511, May1993, and Seshadri and Winters in "Two signaling schemes for improving the error performance of frequency-division-duplex (FDD) transmission systems Transmitter antenna diversity," International Journal of Wireless Information Networks, Vol.1, No.1, 1994 studies transmit diversity. The paper by Wittneben and Seshadri et al. explores transmit diversity from a signal processing point of view.

Space-time codes combine signal processing at the receiver with coding techniques suitable for multiple transmit antennas. See, e.g., V. Tarokh, N. Seshadri, and A.R. Calderbank in "Space-Time Codes For High Data Rate Wireless Communication: Performance Analysis and Code Construction," IEEE Trans. Info. Theory, Vol. 44, No. 2, pp .744-765, March 1998. The space-time scheme is much more efficient than the aforementioned prior art. Specific space-time codes designed for 2-4 transmit antennas perform well in slowly changing fading environments such as indoor transmissions, and have a theoretical outage capacity of 2-3dB. For example, J.Foschini, Jr.And M.J.Gans, "Onlimits of wireless communication in a fading environment, when using multiple antennas," Wireless Personal Communication, Vol.6, No.3, pp.311-335, March 1998 describes interrupt capacity. The code described in the Tarokh et al. paper is about 3-4 times more bandwidth efficient than existing systems. Its most important contribution to performance improvement is diversity, which can be understood as providing multiple copies of the signal transmitted to the receiver, some of which are less susceptible to attenuation. In the paper by Tarokh et al., space-time codes are provided as an optimal trade-off considering constellation size, data rate, diversity gain and trellis complexity.

When the number of transmit antennas is fixed, the decoding complexity (measured, eg, in terms of the number of trellis states in the decoder) increases exponentially with the transmit rate. The decoding complexity can be reduced to some extent by designing a space-time code with a multi-level structure and adopting the multi-level decoding described by Tarokh et al. For a moderate number of transmit antennas (3-6), this approach provides higher data rates while reducing decoding complexity. However, simplified decoding comes at a price. Multi-level decoding is only sub-optimal, partly because the error coefficients are amplified, and this performance penalty means that an alternative technical solution is required to achieve high data rates.

To achieve high data rates in narrowband wireless channels, many antennas are required for both the transmitter and receiver. Consider a wireless communication system with n transmit and m receive antennas, where the sub-channels between each transmit and receive antenna are quasi-static Rayleigh, flat and uncorrelated. If n is fixed, then the capacity increases logarithmically with m. On the other hand, if m is fixed, then intuitively the capacity must reach a state where adding more transmit antennas will not make much difference. Indeed, this can be seen from the mathematical expression for the outage capacity shown in the aforementioned paper by Foschini and Gans. Therefore, it turns out that in the case of one receive antenna, the outage capacity hardly increases when more than 4 transmit antennas are used. Similar evidence shows that using six transmit antennas provides almost all the capacity gain that can be obtained if there are two receive antennas.

If n increases and m ≥ n, then information theory shows that the capacity of the system increases at least linearly as a function of n. Therefore, it makes sense to increase the number of antennas at both the receiver and the transmitter for higher capacity. Simultaneous application of multiple antennas in the transmitter and receiver results in a MIMO system in which the number of degrees of freedom is given by the values of the transmitting and receiving antennas.

Foschini considered such a system in "Layered space-time architecture for wireless communication in a fading environment when using multi-element antennas," Bell Labs Technical Journal, Vol.1, No.2, Autumn 1996. He proposed a multilayer structure that in principle enables a tight lower bound on capacity. If n transmit and n receive antennas are used, then the transmit signal from transmit antenna 1 is considered the desired signal in the receiver, while the signals transmitted from the other transmit antennas are considered interference. Then utilize n receive antennas to perform linear processing to suppress the interfering signal, which provides a diversity gain of 1. Once the signal transmitted from antenna 1 is correctly detected, the signal transmitted from antenna 2 is considered as intended signal, while the signals transmitted from transmitting antennas 3, 4, . . . , n are considered as interference. Since the signal transmitted from antenna 1 has been detected, its effect is subtracted from the signal received by receiver antennas 1-n. Thereafter, detection of signals transmitted from antenna 2 continues linear processing for suppressing interference signals from antennas 3-n. This case provides a diversity gain of 2. Continue to repeat this process until all transmitted signals are detected. Obviously, the worst diversity in this structure is 1. For such systems, long data frames combined with powerful encoding techniques are required to achieve a lower bound on interrupt capacity.

U.S. Patent Application No. 09/114838, filed July 14, 1998, claims priority to U.S. Provisional Patent Application 60/052,689, filed July 16, 1997, which discloses a device that utilizes a combination array Signal processing and channel coding perspectives enable performance enhancements. Specifically, the transmitter's antennas are subdivided into small groups, and information is transmitted from each antenna group using an independent space-time code. In the receiver, an independent space-time code is decoded using a linear array processing technique that suppresses interfering signals emitted by other antenna groups by treating them as interfering signals. The influence of the decoded signal on other received signals is then removed from these received signals. The net result is to provide a simple receiver architecture which provides diversity and coding gain over an uncoded system with a given diversity gain. The combination of array processing at the receiver and multi-transmit antenna coding techniques enables reliable and high-speed communication over wireless channels. An advantage of the packet interference suppression method on the Foschini structure is that the number of receive antennas can be less than the number of transmit antennas.

U.S. Patent Application No. 09/149,163, filed September 4, 1999, claims priority to U.S. Provisional Patent Application 60/052,689, filed July 17, 1997, which discloses an apparatus in which K synchronized terminal devices Signals are sent through N antennas to a base station with M≥K antennas. Improvements are obtained by applying both interference cancellation (IC) and maximum likelihood (ML) decoding. More specifically, space-time block coding is applied when each transmitter uses N transmit antennas, and the signal is received at a receiver with M receive antennas. By utilizing the space-time divisional code structure, when decoding a signal transmitted by a given mobile device, the contributions of K-1 interfering transmitting devices at the receiver are canceled regardless of the number N of transmitting antennas. And the application also discloses an arrangement in which the signal of the first terminal device is decoded first, and when the signals of the remaining K-1 terminal devices are decoded, the resulting decoded signal is used to cancel their The effect on the signal received by the base station antenna. This process is repeated among the remaining K-1 terminal devices.

Performance enhancements can be achieved by combining channel coding with the space-time code principles disclosed in the '163 application. More specifically, since K synchronized terminal devices transmit signals to a base station with M≥K receiving antennas through N antennas, by applying the inner code is a space-time code, and the outer code is a concatenation of conventional channel error correction codes An encoding scheme that enables increased system capacity and enhanced performance. That is, the information symbols are first encoded with a conventional channel code. The channel code is then encoded with a space-time divisional code and transmitted through N antennas. At the receiver, the internal space-time divisional code is used to suppress interference from other co-channel terminals and to make soft decisions on the transmitted symbols. Next, channel decoding makes hard decisions on the transmitted symbols.

The increase in data rate is achieved by efficiently splitting the input data rate into multiple channels, with each channel being transmitted through its own terminal. Viewed another way, information symbols from a sending terminal are divided into L parallel information streams. Stream 1 is encoded with a channel code of rate Rl and then re-encoded with a space-time block encoder with N transmit antennas. The choice of code rate is best as follows:

R ₁ >R ₂ >... >R _L .

Fig. 1 schematically depicts such an apparatus, which includes a base station 30 with 4 antennas, a terminal device 10 with 2 antennas and a terminal device 20 with 2 antennas; and

Figure 2 illustrates that an end device splits an incoming signal into two information streams, and each stream is transmitted through an independent dual-antenna configuration.

The device 10 illustrated in FIG. 1 employs a space-time block encoder 13 followed by a conventional constellation mapper and pulse shaping circuit 16 . The output of circuit 16 is fed to two transmit antennas 11 and 12 . The symbols input to the space-time block encoder are divided into multiple groups, each group consists of two symbols, and in a given symbol period, the two symbols {c ₁ , c ₂ } in each group are transmitted from two antennas simultaneously emission. The signal transmitted from antenna 11 is c ₁ , and the signal transmitted from antenna 12 is c ₂ . In the next symbol period, signal -c ₂ ^* is transmitted from antenna 11 and signal c ₁ ^* is transmitted from antenna 12 .

At receiver 20 the signal is received by antennas 21 and 22 and applied to detector 25 . Channel estimators 23 and 24 act on the input signals of antennas 21 and 24, respectively, in a conventional manner to estimate channel parameters. These estimates are applied to detector 25 . In the mathematical calculations disclosed herein, it is assumed that the channel from each of the two transmit antennas remains constant for two consecutive symbol periods. Right now,

h _i (nT) = h _i ((n+1)T), i = 1, 2 (1)

To determine channel characteristics, the transmitter performs a calibration session during which pilot signals or tones are transmitted. It is these signals that are received during the calibration session and used by the well known channel estimator circuits 23 and 24 .

maximum likelihood detection

The signal received by antenna 21 can be expressed as:

r ₁ =h ₁ c ₁ +h ₂ c ₂ +η ₁ (2) $r_2=-h_1{c}_2^{*}+h_2{c}_1^{*}+{\mathrm{\η}}_2,---\left(3\right)$

Here _r1 and _r2 are signals received in two consecutive symbol periods, _h1 represents the fading channel between transmit antenna 11 and receive antenna 21, _h2 represents the channel between transmit antenna 12 and receive antenna 21, Whereas η ₁ and η ₂ are noise terms, assumed to be compound Gaussian random variables with zero mean and power spectral density N ₀ /2 per dimension. Define vector r=[r ₁ r ₂ ^* ] ^T , c=[c ₁ c ₂ ] ^T , and η=[η ₁ η ₂ ^* ] ^T , formulas (2) and (3) can be rewritten in matrix form as:

r＝H c+η (4)

Here the channel matrix H is defined as: $h=\left[\begin{array}{cc}{h}_1& h_2\\ {h_2}^{*}& {{-h}_2}^{*}\end{array}\right]\·---\left(5\right)$

的符号对

。这可改写为： $\hat{c}=\arg \underset{\hat{c}\∈C}{\min }\left|\right|r-H\·\hat{c}|{|}^2---\left(6\right)$ Vector η is a composite Gaussian random vector with zero mean and covariance N ₀ ·I. Define C as the set of all possible symbol pairs c={c ₁ , c ₂ }, and assume that all symbol pairs are equally probable, so obviously an optimal maximum likelihood (ML) decoder can choose from C that minimizes the expression

pair of symbols

. This can be rewritten as: $\hat{c}=\arg \underset{\hat{c}\∈C}{\min }\left|\right|r-h\&Center\; Dot;\hat{c}|{|}^2---\left(6\right)$

的不确定性Δ_c定义为： ${\mathrm{\Δ}}_c=\left|\right|r-H\·\hat{c}|{|}^2\·---\left(7\right)$ From S.Alamouti's "Space BlockCoding: A simple Transmitter Diversity Scheme for wireless Communications," submitted to IEEE JASC in September 1997, it can be seen that the diversity sequence of the above space-time block code is equivalent to the maximum ratio receiving combination of the two branches ( MRRC) diversity sequence. Alamouti also pointed out that due to the orthogonality of matrix H, this decoding rule is decomposed into two independent decoding rules for _c1 and _c2 . decoding symbols

The uncertainty _Δc of is defined as: ${\mathrm{\Δ}}_c=\left|\right|r-h\&Center\; Dot;\hat{c}|{|}^2\·---\left(7\right)$

。这导致 $\hat{c}=\arg \underset{\hat{c}\∈C}{\min }\left|\right|\tilde{r}-{\left(\right|h_1{|}^2+\left|h_2{|}^2\right)}^2\·\hat{c}|{|}^2\·---\left(9\right)$ 因此通过利用简单的线性组合，公式(9)的解码规则减少为用于c₁和c₂的两个独立、更为简单的解码规则。当利用具有2b个星座点的信令星座时，计算ML解码所需的解码矩阵数从2^2b减少到2×2^b。The maximum likelihood (ML) rule of equation (6) can be simplified by making the channel matrix H orthogonal; ie, H ^* H = (|h ₁ | ² + |h ₂ | ² )I. This results in a modified receive vector, $\tilde{r}=h^{*}r=\left(\right|h_1{|}^2+\left|h_2{|}^2\right)\&Center\; Dot;c+\widetilde{\mathrm{\η}},---\left(8\right)$ here

. this leads to $\hat{c}=\arg \underset{\hat{c}\∈C}{\min }\left|\right|\tilde{r}-{\left(\right|h_1{|}^2+\left|h_2{|}^2\right)}^2\&Center\; Dot;\hat{c}|{|}^2\·---\left(9\right)$ Thus the decoding rule of equation (9) is reduced to two independent, simpler decoding rules for c ₁ and c ₂ by utilizing simple linear combinations. When utilizing a signaling constellation with 2b constellation points, the number of decoding matrices required to compute the ML decoding is reduced from 2 ^2b to 2× ^2b .

When receiver 20 uses M receiver antennas, the reception vector at antenna m is:

r _m =H _m c+η _m , (10)

Here the channel matrix _Hm is defined as: $h_m=\left[\begin{array}{cc}{h}_{1m}& h_{2m}\\ h_{2m}^{*}& {-h}_{1m}\end{array}\right]\&Center\; Dot;---\left(11\right)$

In this case, the optimal ML decoding rule is $\hat{c}=\arg \underset{\hat{c}\∈C}{\min }\underset{m=1}{\overset{m}{\mathrm{\Σ}}}\left|\right|r_m-h_m\&Center\; Dot;\hat{c}|{|}^2,---\left(12\right)$

的相应不确定性Δ_c定义为： ${\mathrm{\Δ}}_c=\underset{m=1}{\overset{M}{\mathrm{\Σ}}}\left|\right|r_m-H_m\·\hat{c}|{|}^2\·---\left(13\right)$ while decoding symbols

The corresponding uncertainty _Δc of is defined as: ${\mathrm{\Δ}}_c=\underset{m=1}{\overset{m}{\mathrm{\Σ}}}\left|\right|r_m-h_m\·\hat{c}|{|}^2\·---\left(13\right)$

Therefore, in the case of M receiving antennas, the decoding rule can be simplified by multiplying the received signal by H _m ^* in advance.

As mentioned above, FIG. 1 shows two terminal equipments 10 and 30. When the two terminal equipments transmit signals simultaneously and in the same channel, the problem to be emphasized is the detection performance of the base station receiver.

In the following notation, g ₁₁ represents the fading channel between the transmitting antenna 31 and the receiving antenna 21, g ₁₂ represents the channel between the antenna 31 and the antenna 22, g ₂₁ represents the channel between the antenna 32 and the antenna 21, and g ₂₂ denotes a channel between antenna 32 and antenna 22 . Likewise {c ₁ , c ₂ } and {s ₁ , s ₂ } denote two symbols transmitted from terminal devices 10 and 30, respectively.

At the receiver 20, the signals _r11 and _r12 received at the receiving antenna 21 during two consecutive symbol periods are:

r ₁₁ =h ₁₁ c ₁ +h ₂₁ c ₂ +g ₁₁ s ₁ +g ₂₁ s ₂ +η ₁₁ (14)

r ₁₂ =-h ₁₁ c ₂ ^* +h ₂₁ c ₁ ^* -g ₁₁ s ₂ ^* +g ₂₁ s ₁ ^* +η ₁₂ (15)

Define r ₁ =[r ₁₁ r ₁₂ ^* ] ^T , c = [c ₁ c ₂ ] ^T , s = [s ₁ s ₂ ] ^T and η ₁ =[η ₁₁ η ₁₂ ^* ] ^T , formulas (14) and (15) can be rewritten in matrix form as:

r ₁ =H ₁ ·c+G ₁ ·s+η ₁ (16)

The channel matrices _H1 and _G1 between the transmitter devices 10 and 30 and the receiving antenna 21 are given by the following equations. $h_1=\begin{array}{c}\left[\begin{array}{cc}{h}_{11}& h_{\mathrm{twenty\; one}}\\ h_{\mathrm{twenty\; one}}^{*}& {-h}_{11}^{*}\end{array}\right],\end{array}$ and $G_1=\left[\begin{array}{cc}{g}_{11}& g_{\mathrm{twenty\; one}}\\ g_{\mathrm{twenty\; one}}^{*}& {-g}_{11}^{*}\end{array}\right]\&Center\; Dot;---\left(5\right)$

Vector η ₁ =[η ₁₁ η ₁₂ ^* ] ^T is a compound Gaussian random vector with zero mean and covariance N ₀ ·I. Similarly, the signals r ₂₁ and r ₂₂ received at the receiving antenna 22 during two consecutive symbol periods are:

r ₂₁ =h ₁₂ c ₁ +h ₂₂ c ₂ +g ₁₂ s ₁ +g ₂₂ s ₂ +η ₂₁ (14)

r ₂₂ =-h ₁₂ c ₂ ^* +h ₂₂ c ₁ ^* -g ₁₂ s ₂ ^* +g ₂₂ s ₁ ^* +η ₂₂ (15)

Define r ₂ =[r ₂₁ r ₂₂ ^* ] ^T and η ₂ =[η ₂₁ η ₂₂ ^* ] ^T in a similar manner, then formulas (16) and (17) can be rewritten as:

r ₂ =H ₂ ·c+G ₂ ·s+η ₂ , (18)

Here the channel matrices _H2 and _G2 are given by: $h_2=\left[\begin{array}{cc}{h}_{12}& h_{\mathrm{twenty\; two}}\\ h_{\mathrm{twenty\; two}}^{*}& {-\mathrm{<figure-callout\; id="12"\; label="h"\; filenames="A9981033800161.png"\; state="\{\{state\}\}">h</figure-callout>}}_{12}^{*}\end{array}\right],$ and $G_2=\left[\begin{array}{cc}{g}_{12}& g_{\mathrm{twenty\; two}}\\ g_{\mathrm{twenty\; two}}^{*}& {-\mathrm{<figure-callout\; id="12"\; label="g"\; filenames="A9981033800161.png"\; state="\{\{state\}\}">g</figure-callout>}}_{12}^{*}\end{array}\right]\&CenterDot;---\left(19\right)$

Formulas (14) and (18) can be combined to obtain the matrix form: $r=\left[\begin{array}{c}{r}_1\\ r_2\end{array}\right]=\left[\begin{array}{cc}{h}_1& G_1\\ h_2& G_2\end{array}\right]\left[\begin{array}{c}c\\ \mathrm{the\; s}\end{array}\right]+\left[\begin{array}{c}{\mathrm{\&eta;}}_1\\ {\mathrm{\&eta;}}_2\end{array}\right].---\left(20\right)$

Minimum Mean Square Error Interference Cancellation (MMSEIC)

When seeking to detect and decode signals {c ₁ , c ₂ } by minimizing the mean square error criterion, the goal is to achieve a linear combination of the received signals such that the mean square error of the detected signals {c ₁ , c ₂ } is minimized. In general, this can be expressed by minimizing an error cost function such as the function: $J(\mathrm{\&alpha;},\mathrm{\&beta;})\left|\right|\underset{i=1}{\overset{4}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_i^{*}{r}_i*({\mathrm{\&beta;}}_i^{*}{c}_1+{\mathrm{\&beta;}}_i^{*}{c}_2)|{|}^2=||{\mathrm{\&alpha;}}^{*}\&CenterDot;r-{\mathrm{\&beta;}}^{*}\&Center\; Dot;c|{|}^2,---\left(\mathrm{twenty\; one}\right)$

Here r=[r ₁ r ₂ r ₃ r ₄ ]T=[r ₁₁ r ₁₂ r ₂₁ r ₂₂ ] ^T .

It may be noted that when both α and β are 0, the minimum is definitely reached, but this is of course not the desired situation. Therefore, set _β1 to 1 or set _β2 to 1.

When _β1 is set to 1, the following minimization criterion can be obtained from formula (40): $J_1({\mathrm{\&alpha;}}_1,{\mathrm{\&beta;}}_1)=\left|\right|\underset{i=1}{\overset{5}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_{1i}^{*}{r}_{1i}-c_1|{|}^2=|\left|{\widetilde{\mathrm{\&alpha;}}}_1^{*}\tilde{r}{-}_1{c}_1\right|{|}^2=J_1\left({\widetilde{\mathrm{\&alpha;}}}_1\right),---\left(\mathrm{twenty\; two}\right)$

here, ${\widetilde{\mathrm{\&alpha;}}}_1=[{\mathrm{\&alpha;}}_{11},{\mathrm{\&alpha;}}_{12},{\mathrm{\&alpha;}}_{13},{\mathrm{\&alpha;}}_{14},-{\mathrm{\&beta;}}_2]=[{\mathrm{\&alpha;}}_1-{\mathrm{\&beta;}}_2],$ and ${\tilde{r}}_1=[r^T{c}_2{]}^T$ . From this it can be seen that ${\tilde{r}}_1=\left[\begin{array}{cc}h& 0\\ 0^T& 1\end{array}\right]\left[\begin{array}{c}\tilde{c}\\ c_2\end{array}\right]+\left[\begin{array}{c}\mathrm{\&eta;}\\ 0\end{array}\right]=R\&CenterDot;d_1+\widetilde{\mathrm{\&eta;}},---\left(\mathrm{twenty\; three}\right)$

Here 0=[0 0 0 0] ^T .

，以便公式(22)中表达式的期望值最小。即选择

以最小化 $E\left\{J_1\left({\widetilde{\mathrm{\&alpha;}}}_1\right)\right\}=E\left\{J_1\left({\widetilde{\mathrm{\&alpha;}}}_1\right)\right\}=E\left\{({\widetilde{\mathrm{\&alpha;}}}_1^{*}{\tilde{r}}_1-c_1){({\widetilde{\mathrm{\&alpha;}}}_1^{*}{\tilde{r}}_1-c_1)}^{*}\right\}---\left(24\right)$ $={\widetilde{\mathrm{\&alpha;}}}_1^{*}E\left\{{\tilde{r}}_1{\tilde{r}}_1^{*}\right\}{\widetilde{\mathrm{\&alpha;}}}_1+E\left\{c_1{c}_1^{*}\right\}-E\left\{{\tilde{r}}_1{c}_1^{*}\right\}{\widetilde{\mathrm{\&alpha;}}}_1^{*}-E\left\{c_1{\tilde{r}}_1\right\}{\widetilde{\mathrm{\&alpha;}}}_1$ we just have to choose

, so that the expected value of the expression in formula (22) is the smallest. ie choose

to minimize $\mathrm{E.}\left\{J_1\left({\widetilde{\mathrm{\&alpha;}}}_1\right)\right\}=\mathrm{E.}\left\{J_1\left({\widetilde{\mathrm{\&alpha;}}}_1\right)\right\}=\mathrm{E.}\left\{({\widetilde{\mathrm{\&alpha;}}}_1^{*}{\tilde{r}}_1-c_1){({\widetilde{\mathrm{\&alpha;}}}_1^{*}{\tilde{r}}_1-c_1)}^{*}\right\}---\left(\mathrm{twenty\; four}\right)$ $={\widetilde{\mathrm{\&alpha;}}}_1^{*}\mathrm{E.}\left\{{\tilde{r}}_1{\tilde{r}}_1^{*}\right\}{\widetilde{\mathrm{\&alpha;}}}_1+\mathrm{E.}\left\{c_1{c}_1^{*}\right\}-\mathrm{E.}\left\{{\tilde{r}}_1{c}_1^{*}\right\}{\widetilde{\mathrm{\&alpha;}}}_1^{*}-\mathrm{E.}\left\{c_1{\tilde{r}}_1\right\}{\widetilde{\mathrm{\&alpha;}}}_1$

的偏导数并将其设为0，这导致 $\left[\begin{array}{cc}M& h_2\\ h_1^{*}& 1\end{array}\right]\left[\begin{array}{c}{\mathrm{\&alpha;}}_1^{*}\\ -{\mathrm{\&beta;}}_2\end{array}\right]=\left[\begin{array}{c}{h}_1\\ 0\end{array}\right],---\left(25\right)$ Pick

and set it to 0, which leads to $\left[\begin{array}{cc}m& h_2\\ h_1^{*}& 1\end{array}\right]\left[\begin{array}{c}{\mathrm{\&alpha;}}_1^{*}\\ -{\mathrm{\&beta;}}_2\end{array}\right]=\left[\begin{array}{c}{h}_1\\ 0\end{array}\right],---\left(25\right)$

here $m={\mathrm{HH}}^{*}+\frac{1}{\mathrm{\&Gamma;}}I$ , Γ is the signal-to-noise ratio, I is a 4×4 identity matrix, _h1 is the first column of H, and _h2 is the second column of H. follow ${\mathrm{\&alpha;}}_1={(m-h_2{h}_2^{*})}^{-1}{h}_1$ and ${\mathrm{\&beta;}}_2^{*}=h_2^{*}{(m-h_2{h}_2^{*})}^{-1}{h}_1---\left(26\right)$

From this it can be seen that ${(m-h_2{h}_2^{*})}^{-1}=m^{-1}+\frac{m^{-1}{h}_2{h}_2^{*}{m}^{-1}}{1-h_2^{*}{m}^{-1}{h}_2},---\left(27\right)$

this gets ${\mathrm{\&beta;}}_2^{*}=\frac{h_2^{*}{m}^{-1}{h}_1}{1-h_2^{*}{m}^{-1}{h}_2}.---\left(28\right)$

From the structure of matrix H, we can easily verify that h ₁ and h ₂ are orthogonal. Using this fact and the structure of the matrix M, it can be found that

β ₂ =0 (29)

α ₁ =M ⁻¹ h ₁ . (30)

Therefore, the MMSE IC technology solution given by equations (29) and (30) will minimize the mean square error in _c1 independent of _c2 . Considering an alternative cost function when _β2 is set to 1, a similar analysis leads to the conclusion:

β ₁ =0 (31)

α ₂ =M ⁻¹ h ₂ (32)

In this case, the MMSE IC technical solution given by equations (31) and (32) will minimize the mean square error in _c2 independent of _c1 . Therefore, according to the formulas (29)-(32), it is obvious that the MMSE interference cancellation for the signal from the terminal device 10 will include two different weight sets α ₁ and α ₂ of c ₁ and c ₂ respectively. The weight to decode the signal from terminal 30 can also be obtained in a similar manner. Thus, the signals from the terminal equipment 10 and 30 can be decoded using a subroutine MMSE.DECODE in the decoder 25 as follows:

(c, _Δc )=MMSE.DECODE(r ₁ ,r ₁ ,H ₁ ,H ₂ ,G ₁ ,G ₂ ,Γ)

{ $\tilde{r}={\left[\begin{array}{cc}{r}_1^T& r_1^T\end{array}\right]}^T$ $\tilde{h}=\left[\begin{array}{cc}{h}_1& G_1\\ h_2& G_2\end{array}\right]$ $m=h{h}^{*}+\frac{1}{\mathrm{\&Gamma;}}I$ $h_1={\left[\begin{array}{cccc}{h}_{11}& h_{\mathrm{twenty\; one}}^{*}& h_{12}& h_{\mathrm{twenty\; two}}^{*}\end{array}\right]}^T=\mathrm{first\; column\; of\; H}$ $h_2={\left[\begin{array}{cccc}{h}_{\mathrm{twenty\; one}}& {-h}_{11}^{*}& h_{\mathrm{twenty\; two}}& {-h}_{\mathrm{twenty\; one}}^{*}\end{array}\right]}^T=\sec \mathrm{ond\; column\; of\; H}$ ${\mathrm{\&alpha;}}_1^{*}=m^{-1}{h}_1,{\mathrm{\&alpha;}}_2^{*}=m^{-1}{h}_2$ $\hat{c}=\underset{{\hat{c}}_1,{\hat{c}}_2\&Element;C}{\arg \min }\left\{\right||{\mathrm{\&alpha;}}_1^{*}\tilde{r}-{\hat{c}}_1|{|}^2+\left|\right|{\mathrm{\&alpha;}}_1^{*}\tilde{r}-{\hat{c}}_2\left|{|}^2\right\}$ ${\mathrm{\&Delta;}}_c=\left|\right|{\mathrm{\&alpha;}}_1^{*}\tilde{r}-{\hat{c}}_1|{|}^2+||{\mathrm{\&alpha;}}_1^{*}\tilde{r}-{\hat{c}}_1|{|}^2$

}

$(\hat{c},\mathrm{\&Delta;})=\mathrm{MMSE}.\mathrm{DECODE}(r_1,r_2,H_1,H_2,G_1,G_2,\mathrm{\&Gamma;})---\left(33\right)$ $(\hat{s},\mathrm{\&Delta;})=\mathrm{MMSE}.\mathrm{DECODE}(r_1,r_2,G_1,G_2,H_1,H_2,\mathrm{\&Gamma;})---\left(34\right)$ Using this subroutine, it is possible to estimate and

$(\hat{c},\mathrm{\&Delta;})=\mathrm{MMSE}.\mathrm{DECODE}(r_1,r_2,h_1,h_2,G_1,G_2,\mathrm{\&Gamma;})---\left(33\right)$ $(\widehat{\mathrm{the\; s}},\mathrm{\&Delta;})=\mathrm{MMSE}.\mathrm{DECODE}(r_1,r_2,G_1,G_2,h_1,h_2,\mathrm{\&Gamma;})---\left(34\right)$

已被正确解码，那么接收机能完全清除终端设备10对接收信号向量r₁和r₂的影响。接收机接着利用X₁和X₂，即清除来自终端设备10信号后的接收信号向量，根据公式(10)中的最优ML解码规则来再解码来自终端设备30的符号。假定来自终端设备10的符号已被正确解码，那么终端设备30的性能将等效于具有2个发射和2个接收天线的设备的性能(等效于4分支MRC分集)。接收机接着重复上述步骤，这是假定来自终端设备30的符号 已利用MMSE.DECODE子程序正确解码。如前所述，接收机清除终端设备30对接收信号向量r₁的影响，并利用y₁和y₂，即清除来自终端设备30信号后的接收信号向量，根据公式(10)中的最优ML解码规则来再解码来自终端设备10的符号

。同样如前所述，假定来自终端设备30的符号已被正确解码，那么终端设备10的性能将等效于具有2个发射和2个接收天线的设备的性能。令Δ₀＝Δ_c0+Δ_s0和Δ₁＝Δ_c1+Δ_s1分别表示 和 以及 和

的总不确定性，接收机比较这两个总不确定性，如果Δ₀＜Δ₁，则选择(

)，否则选择(

)。该两级干扰对消和ML解码算法在下面的伪代码子程序II.MMSE.DECODE中提供。 $(\hat{c},\hat{s})=\mathrm{II}.\mathrm{DECODE}(r_1,r_2,H_1,H_2,G_1,G_2,\mathrm{\&Gamma;})$ Additional improvements can be achieved by applying a two-stage interference cancellation scheme. In this two-stage scheme, the receiver decodes the signals from both terminals using the MMSE.DECODE subroutine published above. Assume that the symbol from the terminal device 10

has been correctly decoded, then the receiver can completely remove the influence of the terminal device 10 on the received signal vectors _r1 and _r2 . The receiver then uses X ₁ and X ₂ , i.e., the received signal vector after clearing the signal from the terminal device 10, to re-decode the symbols from the terminal device 30 according to the optimal ML decoding rule in equation (10) . Assuming that the symbols from terminal device 10 have been correctly decoded, the performance of terminal device 30 will be equivalent to that of a device with 2 transmit and 2 receive antennas (equivalent to 4-branch MRC diversity). The receiver then repeats the above steps, assuming that the symbols from the terminal equipment 30 Correctly decoded using the MMSE.DECODE subroutine. As mentioned above, the receiver removes the influence of the terminal device 30 on the received signal vector r ₁ , and uses y ₁ and y ₂ , that is, the received signal vector after removing the signal from the terminal device 30, according to the optimal ML decoding rules to re-decode symbols from terminal device 10

. Also as before, assuming that the symbols from the terminal device 30 have been correctly decoded, the performance of the terminal device 10 will be equivalent to that of a device with 2 transmit and 2 receive antennas. Let Δ ₀ = Δ _c0 + Δ _s0 and Δ ₁ = Δ _c1 + Δ _s1 to represent and as well as and

The total uncertainty of , the receiver compares these two total uncertainties, if Δ ₀ < Δ ₁ , select (

), otherwise choose (

). The two-stage interference cancellation and ML decoding algorithm is provided in the pseudocode subroutine II.MMSE.DECODE below. $(\hat{c},\widehat{\mathrm{the\; s}})=\mathrm{II}.\mathrm{DECODE}(r_1,r_2,h_1,h_2,G_1,G_2,\mathrm{\&Gamma;})$

{ $({\hat{c}}_0,{\mathrm{\&Delta;}}_{c,0})=\mathrm{MMSE}.\mathrm{DECODE}(r_1,r_2,h_1,h_2,G_1,G_2,\mathrm{\&Gamma;})$ $x_1=r_1-h_1\&CenterDot;{\hat{c}}_0,x_2=r_2-h_2\&Center\; Dot;{\hat{c}}_0$

F(s) = ‖x ₁ -G ₁ ·s‖ ² +‖x ₂ -G ₂ ·s‖ ² ${\widehat{\mathrm{the\; s}}}_0=\underset{\mathrm{the\; s}\&Element;S}{\arg \min }\left(f\left(\mathrm{the\; s}\right)\right),$ Δ _s,0 = F(s) $({\widehat{\mathrm{the\; s}}}_1,{\mathrm{\&Delta;}}_{\mathrm{the\; s},1})=\mathrm{MMSE}.\mathrm{DECODE}(r_1,r_2,G_1,G_2,h_1,h_2,\mathrm{\&Gamma;})$ ${\mathrm{the\; y}}_1=r_1-G_1\&CenterDot;{\widehat{\mathrm{the\; s}}}_1,{\mathrm{the\; y}}_2=r_2-G_2\&Center\; Dot;{\widehat{\mathrm{the\; s}}}_1$

F(c)=‖y ₁ -H ₁ ·c‖ ² +‖y ₁ -H ₂ ·c‖ ² ${\hat{c}}_1=\underset{c\&Element;C}{\arg \min }\left(f\left(c\right)\right),$ Δc _,1 = F(c)

If(Δc _,0 +Δs _,0 )<(Δc _,1 +Δs _,1 ) $(\hat{c},\widehat{\mathrm{the\; s}})=({\hat{c}}_0,{\widehat{\mathrm{the\; s}}}_0)$

Else $(\hat{c},\widehat{\mathrm{the\; s}})=({\hat{c}}_1,{\widehat{\mathrm{the\; s}}}_1)$

}

With the theoretical background disclosed above, we realize that performance enhancements can be achieved by applying space-time block coding to interference cancellation and ML decoding, while also using another coding scheme to overcome channel-induced quality degradation, such as decline. Thus, each transmitter in Figure 1 includes a channel coder (14 and 34 respectively) inserted between the input signal and the space-time block coder of the transmitter. Channel encoders 14 and 24 may apply any conventional channel error correction code (eg trellis or convolutional codes).

In receiver 20, the inner space-time divisional block code is decoded at unit 26 and used to suppress interference from individual co-channel terminals using the MMSE scheme disclosed above. Unit 26 forms two interference cancellation vectors a _il and a _il corresponding to a certain terminal i, and unit 27 forms two decision variables: ${\mathrm{\&xi;}}_1={\mathrm{\&alpha;}}_1^{*}\mathrm{r\; and}{\mathrm{\&xi;}}_2={\mathrm{\&alpha;}}_2^{*}r---\left(35\right)$

These decisions will however be used as soft decisions for the transmitted information symbols and fed to the channel decoder 28 which is a conventional decoder and corresponds to the type of encoding performed in the channel encoders 14 and 34 . Thus, in the arrangement depicted in FIG. 1, the structure of the inner coder is used for interference suppression so that multiple co-channel terminals can operate simultaneously while providing diversity. The output of the inner code space-time decoder forms the input of the outer coder decoder, which determines the transmission information when correcting channel errors.

Figure 2 provides an apparatus for increasing the data rate or throughput of a wireless system. In FIG. 2, the information to be transmitted is split at unit 40 into two information streams. One of the streams is applied to the channel encoder 41 and the other stream is applied to the channel encoder 51 . The output of channel encoder 41 is applied to space-time block encoder 42 , which is then applied to mapper and pulse shaper 53 and antennas 44 and 45 . In general, information symbols from a transmitting terminal are divided into L parallel information streams. Stream l is then encoded with a channel code of rate _Rl and then encoded with a space-time block encoder with N transmit antennas. The coding rates of the two can be the same, but it is more convenient if the coding rate is selected as R ₁ >r ₂ >...> _RL . In this case, symbols transmitted in stream l will be more resistant to channel errors than symbols transmitted in stream u (u>l). The base station receiver is assumed to be equipped with at least L receive antennas. The base station receiver treats each information stream as a distinct user and utilizes the recursive interference cancellation techniques disclosed above, or the techniques disclosed in the aforementioned '163 application. Since the first stream has the minimum code rate R ₁ , it has the best resistance to channel errors and is very likely to be error-free. The receiver then uses the decoded symbols of stream l to cancel the contribution of the first stream to the total received signal while simultaneously decoding the remaining Ll streams. When decoding the remaining L1 streams, the decoder first decodes the signal from the second stream, since it has the best resistance to channel errors among all remaining L1 streams (since its rate R ₂ minimum). The receiver then uses the decoded symbols of the second stream to remove its influence on the received signal. This process repeats until all streams are decoded.

It can be seen that in this case, the system flow can be given by: $\mathrm{\&rho;}=\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{R}_l(1-{\mathrm{FER}}_1),---\left(36\right)$

FER _l is the frame error rate of stream l.

Claims (14)

1. A device comprising: a channel code encoder responsive to an applied input signal, a space-time encoder responsive to the output signal of said channel code encoder; and a modulator responsive to the space-time encoder. 2. The apparatus according to claim 1 , further comprising a pulse shaping circuit and at least two antennas for transmitting the space-time encoded signal generated by said space-time encoder and modulated by said modulator . 3. A transmitter including: a demultiplexer responsive to the applied input signal for generating at least two signal streams, and as many channel-coded/space-time-coded transmitters, each transmitter responsive to a different one of said plurality of signal streams. 4. In a transmitter according to claim 3, each of said channel coding/space-time coding transmitters comprises: A channel coder with code rate R _l , a space-time encoder responsive to the output signal of said channel code encoder, a modulator responsive to said space-time encoder, and At least two antennas for transmitting the space-time encoded signal produced by the space-time encoder, modulated by the modulator, and limited by the pulse shaping circuit. 5. The transmitter according to claim 4, said demultiplexer generates L signal streams, wherein said channel encoders in said L channel coding/space-time coding transmitters generate mutually different code rates R _i i = 1, 2, . . . , L. 6. The transmitter according to claim 4, said demultiplexer generates L signal streams, wherein said channel encoder in said L channel coding/space-time coding transmitters generates a code rate R _i i = 1, 2, ..., L, and R ₁ >R ₂ >... >R _L . 7. The transmitter of claim 1, wherein said channel code encoder performs trellis encoding. 8. The transmitter of claim 1, wherein said channel code encoder performs convolutional encoding. 9. A receiver comprising: a detector for space-time coded signals; and a decoder for decoding the channel code encoded signal embedded in the output signal of said detector. 10. The receiver according to claim 9, wherein said detector employs a MMSE IC decoder. 11. The receiver of claim 9, wherein said detector applies a two-stage algorithm to generate a weight vector for canceling interfering signals from terminals other than the given terminal for which the signal is being detected. 12. The receiver according to claim 11, wherein said two-stage algorithm is: $(\hat{c},\widehat{\mathrm{the\; s}})=\mathrm{II}.\mathrm{DECODE}(r_1,r_2,h_1,h_2,G_1,G_2,\mathrm{\&Gamma;})$ { $(\hat{c},{\mathrm{\&Delta;}}_{c,0})=\mathrm{MMSE}.\mathrm{DECODE}(r_1,r_2,h_1,h_2,G_1,G_2,\mathrm{\&Gamma;})$ $x_1=r_1-h_1\&Center\; Dot;{\hat{c}}_0,x_2=r_2-h_2\&Center\; Dot;{\hat{c}}_0$ F(s) = ‖x ₁ -G ₁ ·s‖ ² +‖x ₂ -G ₂ ·s‖ ² ${\widehat{\mathrm{the\; s}}}_0=\underset{\mathrm{the\; s}\&Element;S}{\arg \min }\left(f\left(\mathrm{the\; s}\right)\right)$ , Δ _{s, 0} = F(s) $({\widehat{\mathrm{the\; s}}}_1,{\mathrm{\&Delta;}}_{\mathrm{the\; s},1})=\mathrm{MMSE}.\mathrm{DECODE}(r_1,r_2,G_1,G_2,h_1,h_2,\mathrm{\&Gamma;})$ ${\mathrm{the\; y}}_1=r_1-G_1\&CenterDot;{\widehat{\mathrm{the\; s}}}_1,{\mathrm{the\; y}}_2=r_2-G_2\&CenterDot;{\widehat{\mathrm{the\; s}}}_1$ F(c)=‖y ₁ -H ₁ ·c‖ ² +‖y ₁ -H ₂ ·c‖ ² ${\hat{c}}_1=\underset{c\&Element;C}{\arg \min }\left(f\left(c\right)\right),$ Δ _c.1 = F(c) If(Δc _,0 +Δs _,0 )<(Δc _,1 +Δs _,1 ) $(\hat{c},\widehat{\mathrm{the\; s}})=({\hat{c}}_0,{\widehat{\mathrm{the\; s}}}_0)$ Else $(\hat{c},\widehat{\mathrm{the\; s}})=({\hat{c}}_1,{\widehat{\mathrm{the\; s}}}_1)$ } 13. The receiver of claim 9, wherein said decoder for decoding a channel code is a trellis code decoder. 14. The receiver of claim 9, wherein the decoder for decoding the channel code is a convolutional decoder.

Images